Re: One Time Pads.
It's a perfectly good question; I don't know why it was downvoted.
The OTP protocol, if done perfectly (truly random pad with a perfectly random distribution, secure key exchange, blah blah blah), produces a ciphertext for which all possible plaintexts of the same length are equally probable. Combine it with some plaintext splitting and padding to add noise to the length signal, or use a protocol such as chaffing to interfere with traffic analysis, and you have as close to a perfectly secure communications channel as possible.1
Perfect Forward Security (PFS) doesn't affect the strength of the protocol, algorithm, or key used to affect the message. It's a feature of the key-exchange protocol. So it's an apples-and-oranges comparison; where OTP is a protocol for encrypting a message, PFS is part of a key-exchange protocol.
PFS simply says "use a unique set of parameters for anonymous key-agreement each time you do key exchange". As an example, look at the evolution of SSL/TLS:
Prior to TLSv1.3, it was very common to use RSA, or Diffie-Hellman (discrete or ECC) with a fixed key, to do key exchange. That is, one side would create a random session key for symmetric encryption of messages, then encrypt it with the peer's public RSA key, and send it to the peer, who would decrypt it using their private key. Or they'd do DH agreement but using a fixed set of parameters.
The problem with this is a well-resourced adversary can save encrypted messages and work on getting that RSA private key (by cracking, or by subverting a machine that has a copy of it, or various other means), or breaking the fixed DH parameters (this was the "WeakDH" part of the Logjam/WeakDH vulnerability a few years ago). Then they could go back through that archive of saved messages and decrypt them.
With PFS, the peers do anonymous DH key agreement to generate a different key each time. If an adversary breaks the key agreement for one session and gets that session key, they only get that one session; they don't get all the others they might have saved.
PFS is such a useful property that all the TLSv1.3 suites use it.
But after key exchange, those suites are doing conventional symmetric encryption (nearly always with AES, though there's also a ChaCha20-Poly1305 suite, and experimental post-quantum ones). They're not using OTPs, and so the entropy in the key is always (in practice) much smaller than the size of the message.
1This is a gloss. "Secure" doesn't mean anything outside the context of a threat model, and there are certainly plausible models where, for example, using a steganographic channel in addition to the other factors would reduce the threat further.